Prediction Markets are a family of Internet–based social computing applications, which use market price to aggregate and reveal information and opinion from dispersed audiences. The considerable complexity of these markets inhibited the full realization of the promise so far. This paper offers the P–MART classification as a tool for organizing the current state of knowledge, aiding the construction of tailored markets, identifying ingredients for Prediction Markets’ success and encouraging research. P–MART is a dual–facet classification of implementations of Prediction Markets describing traders and markets. The proposed classification framework was calibrated by examining a variety of real–world online implementations. A publicly accessible wiki resource accompanies this paper in order to stimulate further research and future expansion of the classification.
2. Prediction Markets: State of the art
3. The P–MART classification scheme
4. Application of the classification
5. Discussion and Conclusion
Prediction Markets are a family of Internet–based social decision support applications which are a subset of a broader concept, “Information Markets”. The broader term describes a variety of ways to exchange information for direct, indirect or no payment. Prediction Markets use the market price to aggregate and reveal information and opinion from dispersed audiences. They build on the collective intelligence of large numbers of people for aggregating information and informing decisions in a variety of areas including marketing, product development, policy issues and more. This novel tool and method for pooling knowledge has seen a surge of academic and business interest leading to a considerable body of research and applications. The current paper aims to organize knowledge about Prediction Markets to serve research, practice and design.
The information generated by Prediction Markets is used by the public, government, non–profits, and businesses to predict future events, support organizational decision processes, elicit and evaluate new ideas and collect public opinion. In Prediction Markets contracts are traded where the underlying asset is a statement to be evaluated or an event to be forecasted. The price of the contract reflects the market estimation as to the probability of an event occurrence or the statement’s chances to be correct. On the due date of the event, the market closes, the price of the contracts that represent the actual outcome are fixed to the predefined amount, and all other contracts related to the event are nullified. The payoffs due to the traders are computed according to their holdings. Examples of typical questions resolved by markets are: Who will win the 2012 Best Picture Academy Award? What will be the sales volume of product X in the last quarter of 2011? Will project Y be completed by December 2011?
The difficulty of engineering systems (political and otherwise) that balance different preferences with different probabilities within large groups of constituents was identified by Zeckhauser (2010) as one of the hardest problems in the social sciences that need solving. By offering support to decision makers and by resembling financial markets Prediction Markets have the potential of evolving into a balancing system, thus generating considerable enthusiasm and concurrent development by commercial enterprises and academic institutions. The rapid and distributed nature of development and evaluation of Prediction Markets has led to a wealth of offerings, however, there is no consensus regarding goals, mechanisms, and applications.
The aim of this paper is to develop a two–facet classification of the implementations of Prediction Markets, and map the relevant research according to this classification. Beyond organizing the knowledge in the field, the purpose of the proposed classification is to extend the understanding of how Prediction Markets work, their commonalities and differences, and identify ingredients for Prediction Markets’ success. Further, the classification can serve as a framework for identifying significant research questions. The proposed classification framework was calibrated by examining real–world implementations of online Prediction Markets that represent the versatility of the tool. A public wiki (http://pm.haifa.ac.il) was set up in order to enable ongoing discussion and continued development and use of the classification. This wiki is a repository of implementations and is organized according to the classification proposed in this paper. It is hoped that with the collaborative effort of the community, it will become a valuable tool and meeting ground for researchers and practitioners.
2. Prediction Markets: State of the art
Underlying the concept of Prediction Markets is the theory of Hayek (1945), who claimed that market prices aggregate the information and the tastes of numerous individuals. The theories of Rational Expectations (Lucas, 1972) and Efficient Markets (Fama, 1970), promoted the two–sided function of markets and prices as mechanisms of information aggregation and dissemination. These functions of markets were later demonstrated in experimental settings (Plott and Sunder, 1982; Plott and Sunder, 1988).
First to implement public online Prediction Markets was the University of Iowa in 1988 with markets which became known as the Iowa Electronic Markets (IEM). These markets were mainly used to predict political events and have since served as a research platform for many studies (Berg and Rietz, 2006). Yet, the concept of aggregating information through a financial markets mechanism awaited a wide scale real–world implementation. The diffusion of the Internet technology facilitated easy deployment of such markets, and access to large and dispersed pools of traders.
In recent years there has been an on-going stream of business organizations that have been experimenting with Prediction Markets in the context of social computing applications. In addition to applications such as sales forecasting (Chen and Plott, 2002), supply chain management (Guo, et al., 2006), and project scheduling (Remidez and Joslin, 2007), there is an increasing interest in the use of Prediction Markets for idea management and innovation (Dahan, et al., 2006; LaComb, et al., 2007).
Prediction Markets have also met with public criticism. In July 2003, the Defense Advanced Research Projects Agency (DARPA) was planning to launch the Policy Analysis Market (PAM), i.e., markets that would aggregate information on global zones of geopolitical risk, mainly the Middle East. Soon after its announcement, the project was dubbed “terrorism futures” and eventually was cancelled before its launch. The failure of this project probably hindered the penetration of Prediction Markets into the public sector, but scholars still advocated that, when used correctly, markets can contribute to policy decision–making (Abramowicz, 2004).
New and creative ways of using Prediction Markets are continuously being evaluated. For example, Prediction Markets can be used for intellectual property (IP) management, by taking advantage of the reward given in markets for early revelation of information (Meloso, et al., 2009). Other developments address agricultural planning and yield (https://www.farmetrics.com/), and following the spread of disease (Polgreen, et al., 2007).
There has been significant growth of scholarship in the field of Prediction Markets. These publications belong to a variety of disciplines spanning business and economics, through politics, law and education (Zhao, et al., 2008). The publication of a dedicated academic journal, the Journal of Prediction Markets, as of 2007 is another sign of community formation and prospect.
In many countries, including the United States, the use of Prediction Markets with real funds is deterred by laws restricting Internet gambling. This is likely to be one of the barriers to wider development and deployment of Prediction Markets. In an attempt to resolve this issue, a group of prominent economists called for lowering these restrictions by allowing the operation of small stakes markets (Arrow, et al., 2008). Other scholars (Cherry and Rogers, 2008) explored the interplay between Prediction Markets, the First Amendment, and the gambling legislation, offering specific recommendations to resolve opposition. In 2008, the U.S. Commodity Futures Trading Committee (CFTC at http://www.cftc.gov/) started a process of re–evaluating regulations pertaining to Prediction Markets. The recent approval of “movie futures” exchanges by the U.S. CFTC (2010) may be the first step in perhaps a wider adoption of Prediction Markets.
A broad overview of Prediction Markets which includes a description of various types of contracts, different market configurations and applications, a discussion of accuracy, and potential pitfalls was outlined by Wolfers and Zitzewitz (2004). This paper presented an optimistic view of the potential of Prediction Markets. Nevertheless, Wolfers and Zitzewitz (2004), and others (Berg and Rietz, 2006; Wolfers and Zitzewitz, 2006), also examined issues for improvement, such as contract and incentive design, price manipulation and biased decision– making. They also noted that market implementations lacked a theoretical underlying model. The key issue, however, was whether Prediction Markets provide better results than other similar tools. Studies that compare the performance of the IEM election prediction markets with public opinion polls exhibited conflicting results. While some studies claimed that markets outperformed other forecasting approaches (Berg, et al., 2008; Van Bruggen, et al., 2010), others (Erikson and Wlezien, 2008) demonstrated that polls are more accurate than markets.
Several attempts to organize the domain of Prediction Markets have been reported in the literature. Spann and Skiera (2003) provided guidelines for the design of corporate Prediction Markets. These guidelines were process oriented and did not elaborate on the platform’s features that needed to be addressed as part of the design process. A taxonomy of Prediction Markets systems, proposed by Ankenbrand and Rudzinski (2005), used the term ‘criteria’ to organize knowledge in the field. These criteria were system features organized into market strategy, market design, information design, market operation and data interpretation. Ankenbrand and Rudzinski’s classification consisted of an extensive list of features and its center of gravity leaned towards the technical details of implementation.
Online Prediction Markets belong to an emerging genre of collective intelligence Web applications. In an attempt to organize this larger emerging field, Malone, et al., (2009) identified a relatively small set of building blocks which can be used to classify different collective intelligence applications. The classification is based on two pairs of related questions. One addresses the application perspective: What is being accomplished (the goal), and How is it being done (the structure/process), and the other deals with user perspective: Who is performing the task (staffing) and Why are they doing it (incentives). Malone, et al.’s collective intelligence classification examined the characteristics common to a broad set of collective intelligence applications and inevitably took a high–level perspective. Their work called for further elaboration at each application level in order to capture specific characteristics.
The classification, which we call the P–MART classification, covers two broad perspectives: the market and the trader. It highlights features that differentiate Prediction Market configurations and were found in academic research to have influence on markets performance in various settings.
3. The P–MART classification scheme
The objective of the proposed P–MART classification is to define a parsimonious yet holistic scheme, which captures the essential determinants in the effective design of Prediction Markets, both technical and human. It groups attributes that influence the operation and performance of the markets into a structured set of classes. The classification builds on Malone, et al.’s collective intelligence classification rationale. The collective intelligence classification identifies a relatively small set of building blocks, named “genes”, which capture the essential characteristics of close to 250 different types of collective intelligence applications. Each of these building blocks answers one of the key questions: What, How, Who, and Why. The Prediction Markets mechanism is defined as one of the How genes, while Prediction Markets as a system are mapped to the Decide gene (What), the Crowd gene, and the Money or Glory genes (Why), depending on the implementation. The P–MART classification complements the high–level building blocks with attributes which are more application specific while maintaining the overall structure of the collective intelligence classification. The “What” and “How” compose the broad market dimension, and “Who” and “Why” describe the trader perspective. Figure 1 depicts these relationships and the classes that they include.
Figure 1: The P–MART classification framework.
Each class contains a set of pre–defined attributes. Markets can then be defined by one attribute from every class, to produce a systematic and structured analysis of the merits and disadvantages of the different implementations. Furthermore, the description of each class is referenced with relevant studies which aid in understanding how to design and customize markets for specific purposes.
The rest of this section provides a brief description of each class, its related attributes and research. This list is not assumed to be exhaustive. It will continue to evolve and expand to accommodate new developments in the field. The public P–MART wiki (http://pm.haifa.ac.il) we constructed can accommodate such developments by any researcher of this domain.
4.1. Market perspective
Objective — the purpose for operating a market may be to:
Aggregate information — collect factual, unknown information which is segmented and/or scattered. Page (2007) identified three cases which can benefit from this kind of market: a. some people know the response to the question and some do not know, but it is not clear who is who; b. parts of the information are distributed between different people and only collectively a correct response can be achieved; and, c. people have a blurry and partial picture of the correct information and an aggregation mechanism can clarify. An extensive survey of laboratory experiments which study the efficiency of the market mechanism for aggregating information can be found in Sunder (1995).
Predict — predict uncertain future events. This is the most widespread objective, which has also lent the markets their name — Prediction Markets. Most of the public markets are used for prediction. The use of markets for this purpose is common in the corporate environment and for public governance (Chen and Plott, 2002; Ortner, 1998; Polgreen, et al., 2007). Opinions regarding the accuracy of market prediction vary. While some studies claimed that markets were more accurate than other forecasting measures (Berg, et al., 2008; Hahn and Tetlock, 2004; Pennock, et al., 2000; Rothschild, 2009), others (Erikson and Wlezien, 2008) demonstrated that polls outperformed the markets.
Elicit opinion — obtain and evaluate preferences or ideas from a large and distributed group of people. Motorola (Burnham, 2009) and General Electric (LaComb, et al., 2007) have implemented such markets in the corporate environment. The use of such markets in the context of innovation has also been studied in controlled laboratory settings (Bothos, et al., 2009; Dahan, et al., 2006).
4.2 Trader perspective
Participation — traders joining the market may be:
Self–selected — join on their own initiative, usually when they perceive themselves knowledgeable or interested in the topic of the market.
Invited — traders are invited by the market owner according to estimated potential contribution and expertise.
While it may be expected that the selection of knowledgeable traders would provide better results, Gruca, et al. (2005) suggest that opening the markets to a wider range of participants increases its accuracy. Trading experience was found to be an important determinant in the performance of information markets (Anderson and Sunder, 1995).
Stake — traders may benefit from market outcome beyond market profits:
No stake in outcome — traders are fully neutral with respect to the outcome of the event which is being traded.
Stakeholders — traders may have vested interest in the outcome of the event traded.
In large public political markets interested traders did cause initial large price changes, but a few marginal traders were effective in reverting prices back to previous levels (Rhode and Strumpf, 2006). In other cases, however, biased traders’ behavior caused market mispricing which prevailed throughout the trading period (Forsythe, et al., 1999). In a corporate setting, traders who submitted ideas to GE’s Imagination Markets and were motivated to raise the price of their own ideas, led to a negative effect on the market idea ranking (Spears, et al., 2009). A prediction market mechanism that does not incentivize undesirable actions, i.e., reducing the expected utility of the market organizer, is proposed by Shi, et al. (2009).
Information — Following Page’s (2007) specification of the cases that can benefit from Prediction Markets, information which is available to the group of traders is defined as:
Perfect — Some people have access to full information but it is not clear who these people are. The ability of markets to disseminate private information was exhibited in early laboratory experiments (Plott and Sunder, 1982) when a critical mass of informed traders is present (Sunder, 1992). In real–world prediction markets (Rhode and Strumpf, 2006) ex post analysis also indicates that insiders’ information is reflected in market price.
Complete — Each trader only holds uncertain or incomplete information. Combining the information all traders have access to forms a complete picture of the subject of trade. The ability of markets to aggregate information under such conditions was shown by Plott and Sunder (1988).
Incomplete — Available information, private or public, is blurry or contains a variable level of uncertainty. This condition is typical to real–world scenarios, mainly prediction or opinion markets. When information is incomplete the ability of markets to aggregate information depends on their design (Sunder, 1995). A case study of scientific hypotheses evaluation shows that the combination of private and public information provides best results (Almenberg, et al., 2009).
Incentive Scheme — The criteria for rewarding the traders may be:
Performance — Payoffs are directly tied to the performance of the trader, i.e., profits from trading.
Performance rank — Reward is associated with the rank of the trader in the descending list of performance.
Participation — Reward is tied to participation. Can be associated with participation level or drawn by lottery.
Contribution — Participants are rewarded for their contribution to the marketplace, e.g., creation of new markets.
Research has shown that when performance–based incentives are used, traders trade more aggressively, willing to take more risk, but exhibit lower than randomly expected accuracy. The use of a rank–based scheme, however, leads to significantly better performance (Luckner, 2006).
Reward — can be of the following form:
Money — a cash payment.
Gift — a material prize that is not monetary, e.g., a book, movie tickets.
Bonus securities — reward of additional units of securities.
Glory — increased recognition and status, rewards the participant’s self–esteem.
Love — no exogenous reward, only pleasure and satisfaction and the gratifying value of the activity itself.
Due to budgetary and regulatory constraints, many of the Prediction Markets reward the traders with non–economic or low–value rewards. Prediction Markets exist within the Internet cultural milieu, where intrinsic and social motivations play an important role in participation and contribution (Benkler, 2006). In this environment, Raban (2008) has demonstrated that in a market for information which provides monetary as well as social rewards, the social incentives are those that induce persistent participation by the contributor.
4. Application of the classification
To examine the usefulness of the P–MART classification we applied it to 10 real–world implementations that represent different types and configurations of Prediction Markets. Tables 1 and 2 summarize the attributes for the 10 selected markets at the market and trader levels.
Table 1: Summary of cases classification from the market perspective.
Table 2: Summary of cases classification from the trader’s perspective.
To illustrate the use of P–MART each market was described by a structured narrative that incorporates attributes from all classes. An example of such a narrative, where the classification attributes are marked in italiized text, is hereby presented below:
General Electric’s Imagination Markets
(LaComb, et al., 2007; Spears, et al., 2009)
The objective of the GE’s Imagination Markets was to elicit opinion in order to promote technology innovation within the company. The marketplace was set in a corporate environment, the markets were large (85 participants) and open for a variable duration. The marketplace employed the Continuous Double Auction (CDA) market model, and operated with play money. The final dividends were calculated based on the last five days Volume Weighted Average Price (VWAP). The traders received initial capital endowment. On–platform interaction was enabled through blogs, and off–platform interaction was possible through casual meetings and organized trading parties.
All members of the Computing and Decision Sciences Technology Center were invited to trade. Some of the traders were also idea contributors, and consequently stakeholders. The traders were experienced in the field of the ideas but could not be certain of their future success, therefore possessing incomplete information. The contribution incentive was monetary, while gifts were awarded to traders on performance–rank and lottery drawn participation basis.
Additional narratives displaying the usefulness of the P–MART classification are available in the P–MART wiki.
5. Discussion and conclusion
The field of Prediction Markets is still young and as such will inevitably undergo a process of winnowing and refinement. The purpose of the proposed framework is to provide a common language and comparative frame of reference to allow discussion of commonalities and differences, and to map the current state of the knowledge in the field.
For example, looking at Table 1 that is devoted to market design we see that Inkling Markets and Intrade are similar in terms of their objectives, topics, settings, duration and market size. They differ in their market models, security types, trading funds and currency, and the availability of communication. In other words, in this example the main differences are the engine supporting the trading, the incentive scheme and the social community which is built around the trade. Other comparisons will surface other differences. Looking across Table 1 we can quickly identify that CDA is the most popular market model currently and that most markets provide an initial endowment to their traders.
Table 2 (trader characteristics) is smaller in size than Table 1 and seems to display more uniformity across markets, implying that possibly at the current state of Prediction Markets development most efforts were devoted to market and mechanism design and less to trader characteristics and preferences. Obviously, this aspect emerges as a field for further research and development. Beyond the brief tabular description, using the P–MART’s vocabulary we can develop structured Prediction Markets narratives that deliver the message briefly and effectively.
The P–MART classification depicts the static attributes of Prediction Markets implementations. A classification of these attributes is important for the process of market design and the understanding of design consideration which may affect market performance. Its usefulness appears both at the single market level by description and narrative, and at a bird’s eye overview given in summary tables such as the ones presented here. It should be noted however that such a classification does not address factors which are the result of market dynamics. Traders’ behavior in this particular environment is a recognized determinant in the quality of market outcome, whether behavior is conscious such as price manipulation, or unconscious such as cognitive decision biases. These aspects have generated an increasing number of studies in this area. A classification of the dynamic factors which affect Prediction Markets performance calls for further research.
The success of Prediction Markets in providing accurate and reliable outcomes will determine their future viability. Market performance is not determined by any single factor. Only a holistic view, which takes into account multiple static and dynamic market features, supported by a foundation of theory, can provide the understanding of this complex mechanism. The contribution of the P–MART classification to this process is by placing a coherent structure over the static market attributes and the related research.
Collective intelligence is not just a characteristic of Prediction Markets, it can be used to understand and learn more about them. As an extension of this analysis, the classification was used to construct a public wiki in order to stimulate collaborative creation of a comprehensive database of available implementations as well as further develop the classification itself (see Figure 2). The field of Prediction Markets is young and at a stage of fast evolution. It calls for a dynamic and structured resource, which complements the publication of a fixed, print resource. Combining these two forms of knowledge dissemination complies with new approaches to academic publication which are explored in fast-growing research disciplines (Hoffmann, 2008). Such an open and collaborative platform will facilitate an agile process of knowledge aggregation in the field. Generating a broad and dynamic classification supported by a collaborative wiki will serve the purpose of strengthening the professional Prediction Markets community.
Figure 2: A screenshot from the P–MART wiki depicting the IEM structured narrative.
About the authors
Dorit Geifman is a Ph.D. candidate at the Graduate School of Management, University of Haifa, and is active in the Sagy Center for Internet Research (http://infosoc.haifa.ac.il). She holds a M.Sc. degree in computer science, a B.Sc. degree with distinction in mathematics and history and philosophy of science. Her research interest lies in the field of collective intelligence platforms, and specifically in judgment biases in Information Markets.
E–mail: dgeifman [at] gsb [dot] haifa [dot] ac [dot] il
Daphne R. Raban is Senior Lecturer at the Graduate School of Management, University of Haifa. She is active in the Sagy Center for Internet Research. Her broad area of research interest is the value of information including topics such as information markets, economics of information goods, information/knowledge sharing, the interplay between social and economic incentives, and games and simulations. In collaboration with IBM HRL she conducts social computing research.
E–mail: draban [at] gsb [dot] haifa [dot] ac [dot] il
Sheizaf Rafaeli is the Director of the Sagy Center for Internet Research and Head of the Graduate School of Management, University of Haifa. He is a graduate of the University of Haifa, Ohio State University, and Stanford University. He has held faculty positions at these three institutions, as well as half a dozen other universities. Sheizaf is interested in computers as media, and has been researching and developing in this field since the 1970s.
E–mail: sheizaf [at] rafaeli [dot] net
M. Abramowicz, 2004. “Information markets, administrative decisionmaking, and predictive cost–benefit analysis,” University of Chicago Law Review, volume 71, number 3, pp. 933–1,020.
J. Almenberg, K. Kittlitz and T. Pfeiffer, 2009. “An experiment on prediction markets in science.” PLoS One, volume 4, number 12, at http://www.plosone.org/, accessed 28 June 2011.
M.J. Anderson and S. Sunder, 1995. “Professional traders as intuitive Bayesians,” Organizational Behavior and Human Decision Processes, volume 64, number 2, pp. 185–202.http://dx.doi.org/10.1006/obhd.1995.1099
B.H. Ankenbrand and C. Rudzinski, 2005. “Description and analysis of information markets,” working paper, Universität Witten/Herdecke.
K.J. Arrow, R. Forsythe, M. Gorham, R. Hahn, R. Hanson, J.O. Ledyard, S. Levmore, R. Litan, P. Milgrom, F.D. Nelson, G.R. Neumann, M. Ottaviani, T.C. Schelling, R.J. Shiller, V.L. Smith, E. Snowberg, C.R. Sunstein, P.C. Tetlock, P.E. Tetlock, H.R. Varian, J. Wolfers and E. Zitzewitz, 2008. ”The promise of prediction markets,” Science, volume 320, number 5878, pp. 877–878.http://dx.doi.org/10.1126/science.1157679
Y. Benkler, 2006. The wealth of networks: How social production transforms markets and freedom. Mew Haven, Conn.: Yale University Press.
J.E. Berg and T.A. Rietz. 2006. “The Iowa electronic markets: Stylized facts and open issues,” In: R.W. Hahn and P.C. Tetlock (editors). Information markets: A new way of making decisions. Washington, D.C.: AEI–Brookings Joint Center for Regulatory Studies, pp. 142–169, and at http://tippie.uiowa.edu/iem/research/abstracts/bergrietz_2006.html, accessed 28 June 2011.
J.E. Berg, F.D. Nelson and T.A. Rietz. 2008. “Prediction market accuracy in the long run,” International Journal of Forecasting, volume 24, number 2, pp. 283–298, and at http://tippie.uiowa.edu/iem/research/longrunaccuracy/, accessed 28 June 2011.
E. Bothos, D. Apostolou and G. Mentzas. 2009. “Collective intelligence for idea management with Internet–based information aggregation markets,” Internet Research, volume 19, number 1, pp. 26–41.http://dx.doi.org/10.1108/10662240910927803
K. Burnham, 2009. “How Motorola uses prediction markets to choose innovations,” ITWorld (4 May), at http://www.itworld.com/mobile-amp-wireless/67371/how-motorola-uses-prediction-markets-choose-innovations, accessed 12 July 2010.
K.–Y. Chen and C.R. Plott. 2002. “Information aggregation mechanisms: Concept, design and implementation for a sales forecasting problem,” California Institute of Technology, Division of the Humanities and Social Sciences, Social Science Working Paper, number 1131 (March), at http://www.hss.caltech.edu/SSPapers/wp1131.pdf, accessed 28 June 2011.
M.A. Cherry and R.L. Rogers, 2008. “Prediction markets and the First Amendment,” University of Illinois Law Review, volume 2008, number 3, pp. 833–879, and at http://illinoislawreview.org/article/prediction-markets-and-the-first-amendment/, accessed 28 June 2011.
B. Cowgill, J. Wolfers and E. Zitzewitz, 2008. “Using prediction markets to track information flows: Evidence from Google,” at http://www.bocowgill.com/GooglePredictionMarketPaper.pdf, accessed 28 June 2011.
E. Dahan, A.W. Lo, T. Poggio, N. Chan and A. Kim, 2006. “Securities trading of concepts (STOC),” at http://www.anderson.ucla.edu/faculty/ely.dahan/content/chan_dahan_lopoggio.pdf, accessed 28 June 2011.
R.S. Erikson and C. Wlezien, 2008. “Are political markets really superior to polls as election predictors?” Public Opinion Quarterly, volume 72, number 2, pp. 190–215.http://dx.doi.org/10.1093/poq/nfn010
E.F. Fama, 1970. “Efficient capital markets: A review of theory and empirical work,” Journal of Finance, volume 25, number 2, pp. 383–417.http://dx.doi.org/10.2307/2325486
R. Forsythe, T.A. Rietz and T.W. Ross. 1999. “Wishes, expectations and actions: A survey on price formation in election stock markets,” Journal of Economic Behavior & Organization, volume 39, number 1, pp. 83–110.http://dx.doi.org/10.1016/S0167-2681(99)00027-X
T.S. Gruca, J.E. Berg and M. Cipriano, 2005. “Consensus and differences of opinion in electronic prediction markets,” Electronic Markets, volume 15, number 1, pp. 13–22, and at http://www.electronicmarkets.org/articles/issues/volume-15/volume-15-issue-1, accessed 28 June 2011.
Z. Guo, F. Fang and A.B. Whinston, 2006. “Supply chain information sharing in a macro prediction market,“ Decision Support Systems, volume 42, number 3, pp. 1,944–1,958.
R.W. Hahn and P.C. Tetlock, 2004. “Harnessing the power of information: A new approach to economic development,” AEI–Brookings Joint Center for Regulatory Studies, Working Paper, number 04–21 (September), at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=641444, accessed 28 June 2011.
F.A. Hayek, 1945. “The use of knowledge in society,” American Economic Review, volume 35, number 4, pp. 519–530.
R. Hoffmann, 2008. “A wiki for the life sciences where authorship matters,” Nature Genetics, volume 40, pp. 1,047–1,051, and at http://www.nature.com/ng/journal/v40/n9/full/ng.f.217.html, accessed 28 June 2011.
C.A. LaComb, J.A. Barnett and Q. Pan, 2007. “The imagination market,” Information Systems Frontiers, volume 9, numbers 2–3, pp. 245–256.
R.E. Lucas, Jr., 1972. “Expectations and the neutrality of money,” Journal of Economic Theory, volume 4, number 2, pp. 103–124.http://dx.doi.org/10.1016/0022-0531(72)90142-1
S. Luckner, 2006. “Prediction markets: How do incentive schemes affect prediction accuracy?” Dagstuhl Seminar Proceedings, at http://drops.dagstuhl.de/volltexte/2007/1002/pdf/06461.LucknerStefan.Paper.1002.pdf, accessed 28 June 2011.
T.W. Malone, R. Laubacher and C. Dellarocas, 2009. “Harnessing crowds: Mapping the genome of collective intelligence,” MIT Center for Collective Intelligence, number 2009–01, at http://www.realtechsupport.org/UB/MRIII/papers/CollectiveIntelligence/MIT_CollectiveIntelligence2009.pdf, accessed 28 June 2011.
D. Meloso, J. Copic and P. Bossaerts, 2009. “Promoting intellectual discovery: Patents Versus markets,” Science, volume 323, number 5919, pp. 1,335–1,339.
G. Ortner, 1998. “Forecasting markets? An industrial application,” Technical University of Vienna, Department of Managerial Economics and Industrial Organization, Working paper.
S.E. Page, 2007. The difference: How the power of diversity creates better groups, firms, schools, and societies. Princeton, N.J.: Princeton University Press.
D.M. Pennock, S. Lawrence, C.L. Giles and F.A. Nielsen. 2000. “The power of play: Efficiency and forecast accuracy in Web market games,” NEC Research Institute, Technical Report, number 2000–168, at http://artificialmarkets.com/am/pennock-neci-tr-2000-168/, accessed 28 June 2011.
C.R. Plott and S. Sunder, 1988. “Rational expectations and the aggregation of diverse information in laboratory security markets,” Econometrica, volume 56, number 5, pp. 1,085–1,118.
C.R. Plott and S. Sunder, 1982. “Efficiency of experimental security markets with insider information: An application of rational–expectations models,” Journal of Political Economy, volume 90, number 4, pp. 663–698.http://dx.doi.org/10.1086/261084
P.M. Polgreen, F.D. Nelson and G.R. Neumann, 2007. “Use of prediction markets to forecast infectious disease activity,” Clinical Infectious Diseases, volume 44, number 2, pp. 272–279.http://dx.doi.org/10.1086/510427
D.R. Raban, 2008. “The incentive structure in an online information market,” Journal of American Society of Information Science and Technology, volume 59, number 14, pp. 2,284–2,295.
H. Remidez Jr. and C. Joslin, 2007. “Using prediction markets to support IT project management — Research in progress,” International Research Workshop on IT Project Management 2007, at http://aisel.aisnet.org/irwitpm2007/10/, accessed 28 June 2011.
P.W. Rhode and K.S. Strumpf, 2006. “Manipulating political stock markets: A field experiment and a century of observational data,” at http://www.unc.edu/~cigar/papers/ManipNBER.pdf, accessed 28 June 2011.
D. Rothschild, 2009. “Forecasting elections: Comparing prediction markets, polls, and their biases,” Public Opinion Quarterly, volume 73, number 5, pp. 895–916.http://dx.doi.org/10.1093/poq/nfp082
P. Shi, V. Conitzer and M. Guo, 2009. “Prediction mechanisms that do not incentivize undesirable actions,” Proceedings of the Workshop on Internet & Network Economics (WINE) 2009, Lecture Notes in Computer Science, number 5929, pp. 89–100.
M. Spann and B. Skiera. 2003. “Internet–based virtual stock markets for business forecasting,” Management Science, volume 49, number 10, pp. 1,310–1,326.http://dx.doi.org/10.1287/mnsc.49.10.1310.17314
B. Spears, C. LaComb, J. Interrante, J. Barnett and D. Senturk–Dogonaksoy, 2009. “Examining trader behavior in idea markets: An implementation of GE’s imagination markets,” Journal of Prediction Markets, volume 3, number 1, pp. 17–39.
S. Sunder, 1995. “Experimental asset markets: A survey,” In: J.H. Kagel and A.E. Roth (editors). Handbook of experimental economics. Princeton, N.J.: Princeton University Press, pp. 445–500.
S. Sunder, 1992. “Market for information: Experimental evidence,” Econometrica, volume 60, number 3, pp. 667–695.http://dx.doi.org/10.2307/2951588
U.S. Commodity Futures Trading Commission (CFTC), 2010. “CFTC approves box office receipt contracts submitted by media derivatives,” Press release PR5834–10 (14 June), at http://www.cftc.gov/PressRoom/PressReleases/pr5834-10.html, accessed 12 July 2010.
G.H. Van Bruggen, M. Spann, G.L. Lilien and B. Skiera, 2010. “Prediction markets as institutional forecasting support systems,” Decision Support Systems, volume 49, number 4, pp. 404–416.http://dx.doi.org/10.1016/j.dss.2010.05.002
J. Wolfers and E. Zitzewitz, 2006. “Five open questions about prediction markets,” NBER Working Paper, number 12060, at http://www.nber.org/papers/w12060, accessed 28 June 2011.
J. Wolfers and E. Zitzewitz, 2004. “Prediction markets,” Journal of Economic Perspectives, volume 18, number 2, pp. 107–126.http://dx.doi.org/10.1257/0895330041371321
R. Zeckhauser, 2010. “Solving the social sciences’ hard problems,” at http://harvardmagazine.com/dispatches/social-sciences-hard-problems, accessed 12 October 2010.
S.J. Zhao, C. Wagner and H. Chen, 2008. “Review of prediction market research: Guidelines for information systems research,” PACIS 2008 Proceedings, at hhttp://aisel.aisnet.org/pacis2008/193/, accessed 28 June 2011.
Received 13 October 2010; accepted 10 June 2011.
Copyright © 2011, First Monday.
Copyright © 2011, Dorit Geifman, Daphne R. Raban and Sheizaf Rafaeli.
P–MART: Towards a classification of online prediction markets
by Dorit Geifman, Daphne R. Raban and Sheizaf Rafaeli.
First Monday, Volume 16, Number 7 - 4 July 2011